Patchwork ppc: Avoid decrementer related kvm exits

login
register
mail settings
Submitter David Gibson
Date Oct. 17, 2011, 5:26 a.m.
Message ID <1318829177-30777-1-git-send-email-david@gibson.dropbear.id.au>
Download mbox | patch
Permalink /patch/120091/
State New
Headers show

Comments

David Gibson - Oct. 17, 2011, 5:26 a.m.
In __cpu_ppc_store_decr(), we set up a regular timer used to trigger
decrementer interrupts.  This is necessary to implement the decrementer
properly under TCG, but is unnecessary under KVM (true for both Book3S-PR
and Book3S-HV KVM variants), because the kernel handles generating and
delivering decrementer exceptions.

Under kvm, in fact, the timer causes expensive and unnecessary exits from
kvm to qemu.  This patch, therefore, disables setting the timer when kvm
is in use.

Signed-off-by: Anton Blanchard <anton@au1.ibm.com>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
---
 hw/ppc.c |    6 ++++++
 1 files changed, 6 insertions(+), 0 deletions(-)
Alexander Graf - Oct. 20, 2011, 5:08 p.m.
On 16.10.2011, at 22:26, David Gibson wrote:

> In __cpu_ppc_store_decr(), we set up a regular timer used to trigger
> decrementer interrupts.  This is necessary to implement the decrementer
> properly under TCG, but is unnecessary under KVM (true for both Book3S-PR
> and Book3S-HV KVM variants), because the kernel handles generating and
> delivering decrementer exceptions.
> 
> Under kvm, in fact, the timer causes expensive and unnecessary exits from
> kvm to qemu.  This patch, therefore, disables setting the timer when kvm
> is in use.
> 
> Signed-off-by: Anton Blanchard <anton@au1.ibm.com>
> Signed-off-by: David Gibson <david@gibson.dropbear.id.au>

Thanks, applied to ppc-next.


Alex
David Gibson - Oct. 21, 2011, 1:13 a.m.
On Thu, Oct 20, 2011 at 10:08:36AM -0700, Alexander Graf wrote:
> 
> On 16.10.2011, at 22:26, David Gibson wrote:
> 
> > In __cpu_ppc_store_decr(), we set up a regular timer used to trigger
> > decrementer interrupts.  This is necessary to implement the decrementer
> > properly under TCG, but is unnecessary under KVM (true for both Book3S-PR
> > and Book3S-HV KVM variants), because the kernel handles generating and
> > delivering decrementer exceptions.
> > 
> > Under kvm, in fact, the timer causes expensive and unnecessary exits from
> > kvm to qemu.  This patch, therefore, disables setting the timer when kvm
> > is in use.
> > 
> > Signed-off-by: Anton Blanchard <anton@au1.ibm.com>
> > Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
> 
> Thanks, applied to ppc-next.

I think you've forgotten to push out the new ppc-next again..

Patch

diff --git a/hw/ppc.c b/hw/ppc.c
index 25b59dd..d29af0b 100644
--- a/hw/ppc.c
+++ b/hw/ppc.c
@@ -662,6 +662,12 @@  static void __cpu_ppc_store_decr (CPUState *env, uint64_t *nextp,
 
     LOG_TB("%s: %08" PRIx32 " => %08" PRIx32 "\n", __func__,
                 decr, value);
+
+    if (kvm_enabled()) {
+        /* KVM handles decrementer exceptions, we don't need our own timer */
+        return;
+    }
+
     now = qemu_get_clock_ns(vm_clock);
     next = now + muldiv64(value, get_ticks_per_sec(), tb_env->decr_freq);
     if (is_excp) {